Nine healthy and well-trained cyclists (5 men, 4 women, mean age 26.8 ± 9 yr, VO2max 61.9 ± 7.7 mL.kg-1.min-1) completed both experimental time-trials, which was previously approved by the University of Otago Human Ethics Committee (Dunedin, Otago, New Zealand) and complied with the Helsinki Declaration. Each participant provided written informed consent prior to beginning the study.
Participants completed a double-blinded randomised crossover study, consisting of a pre-testing session, familiarisation trial, and two experimental time trials separated by 7 – 14 days during which time participants were asked to do minimal training. The pre-testing session involved a graded VO2max test on a stationary cycle ergometer (Monark 915E, Varberg, Sweden), with gaseous exchange measured on a Metalyser 3B (Cortex, Biophysik GmbH, Leipzig, Germany). The test began with a 5 min warm-up at a light intensity. Workload then increased every 3 min, with heart rate (Polar 310, Polar, Oulu, Finland) and Rating of Perceived Exertion (RPE) on the Borg scale  measured in the last 30 s of each stage. VO2max was determined when heart rate was consistently within 10 beats of the calculated maximum, the RPE exceeded 19 on the Borg scale , the participant was unable to maintain an RPM above 70 rpm, or the RER was consistently above 1.10. A level 1 trained International Society of Advanced Kinanthropometry (ISAK) anthropometrist also performed an anthropometric assessment during this initial visit, collecting a ‘restricted profile’ as described by ISAK . The ‘restricted profile’ includes a sum of 8 skinfolds, waist and hip girth, body mass, and height.
The familiarisation and experimental time-trials were a 72 km ‘out and back’ road cycle over hilly terrain in Dunedin, New Zealand, in which participants used their own road cycle. Pre-trial diets were replicated from a one day estimated diet record kept in the day preceding the familiarisation trial. Likewise, participants arrived to all trials fasting, and a standardised pre-race breakfast (3152 ± 1847 kJ; 27 ± 11 g protein; 112 ± 49 g CHO; 11 ± 12 g total fat) was provided to participants one hour before the time-trial started.
Measurements took place immediately pre and post time-trial, and then once more after a post-race meal approximately 40 min from finishing, all samples were obtained in the sitting position. A 1 mL capillary blood sample was collected after appropriate cleaning with an alcohol swab, via fingerprick, (Unistick 3 extra lancet, Owen Mumford, Oxford, United Kingdom) and analysed using an i-STAT point of care analyser with a CG8+ cartridge (Abbott Point of Care Inc, Illinois, USA). This provides measures of sodium, haematocrit, and haemoglobin from these measures plasma volume was calculated using the equations of Dill and Costill . Participants were then asked to provide a urine sample in private, which was collected in a 20 mL sealed, sterile plastic tube (Techno Plas, South Australia, Australia) and stored at 4°C until laboratory analysis. A 100 mm visual analogue scale subjective questionnaire regarding thirst, gastrointestinal distress, as previously utilised by Rolls et al.  was also completed by participants both pre and post time-trial. Body mass was measured on electronic scales to the nearest 0.1 kg (Tanita-Wedderburn TBF-310, Illinois, USA) in minimal clothing. Finally, sweat patches (Tagaderm patch + pad, 3 M, Loughborough, UK) were applied to the upper back, forearm, chest and mid thigh on the right-hand side of the body which was first cleaned with deionised water and dried. The patches remained in place throughout the trial. Immediately following the time-trial the patches were removed with sterile tweezers and stored in a 30 mL sealed, sterile plastic tube (Techno plas, South Australia, Australia) at 4°C.
The time-trial course was on a sheltered, this limited the exposure to the wind which was also minimised by starting the time-trials early in the morning a time when wind is minimal, but hilly cycle route in Dunedin, New Zealand, with a total of 1 556 m in elevation gained in the 72 km. Cyclists were given a coded, clear zip-lock bag each containing 15 clear capsules with either 233 mg sodium chloride, or an identical corn flour placebo. Participants were instructed to consume three capsules for every hour, which equated to 700 mg NaCl.h-1, consistent with doses used in previous trials [2, 11], and recommended by Zapf et al. . Water and ‘Jet Plane’ lollies (Pascall, Auckland, New Zealand) could be consumed ad libitum during the trial but the weights consumed were recorded to the nearest 0.1 g (Salter Vista Electronic Scales, England). Participants were fitted with a Global Positioning System (GPS) stopwatch and heart rate monitor (Garmin Forerunner 110, Garmin, Kansas, USA) for the time-trial. Outside air temperature, humidity, and weather were recorded every 15 min during the time-trials using a WS9623 Wireless 868 MHz Weather Station (La Crosse Technology, France).
Performance was assessed via overall time to complete the time-trial. The cyclists’ uphill time splits were also used as a measure of performance to account for any variation in skill in descending the hills. Plasma [Na+] (mmol.L-1), haematocrit, and blood glucose values (mmol.L-1) were analysed via the i-STAT point of care analyser (Abbott Point of Care Inc, Illinois, USA) and recorded in the field. Sweat sodium and chloride concentration (sweat [Na+], sweat [Cl-]) was analysed in small batches through a Cobas C311 module (Roche Diagnostics, Basel, Switzerland) using the Ion Selective Electrode (ISE) technique (mean CV = 2.01 ± 1.59%). Sweat sodium concentrations were then extrapolated to whole body sweat sodium losses using the calculations of Patterson et al. . To ensure contamination of the patches nor leaching from the skin had not occurred sweat potassium was measured and all samples were within the normal range . Urine osmolality was measured via freezing point depression (Osomat 030, Genotec GmbH, Baden-Wurttemberg, Germany), to indicate hydration status. Subjective feelings of thirst were indicated on a 100 mm visual analogue scale, which was used as a rating from 0 (not thirsty at all) to 100 (extremely thirsty) .
Statistical analyses were performed using Stata Version 11.2 (StataCorp, Texas, USA). Normality of the data was evaluated using a Shapiro-Wilks test, and difference in variance was assessed by two-group variance comparison tests before all comparisons.
Multivariate regression was used to assess the effect of sodium supplements on exercise performance and plasma [Na+]. Differences in overall time and uphill time were compared whilst controlling for temperature and weather (wet or dry road). The difference in absolute (mmol.L-1) and relative (%) plasma [Na+] change was analysed controlling for average heart-rate. A paired t-test was also used to investigate differences in plasma [Na+] from pre-race to post-race within each intervention.
Urine and sweat concentrations were well distributed and the absolute (mmol.L-1) and relative (%) change in electrolytes in each were analysed using a Student’s t-test. Changes in body mass, haematocrit, plasma volume change and fluid intake were assessed using multivariate regression controlling for mean heart rate and temperature.
Statistical significance was set at p ≤ 0.05. If a relationship was close to statistical significance, a Cohen’s d effect size was also calculated. Data is reported as mean ± standard deviation (SD).